[wdmmg-discuss] CRA 2010: progress report [was: CRA 2010: description and questions]

Wed Aug 18 22:57:50 UTC 2010

> (disclaimer: I haven't looked yet at the 2010 data
> but am assuming it isn't that different from the
> 2009 data as I gather from the discussion so far)

It is a bit different - there are two tables this year. One classifies
items by COFOG2 but not regions (only by nation); the other has
regions but not COFOG2. See Tables 9 and 10 at:
http://www.hm-treasury.gov.uk/pesa2010_section4.htm

Crucially, this means we have some COFOG2 information for local
authority items, unlike CRA2009. Sorry, I should probably have
highlighted that.

Most of the rows are duplicates, and can be identified by the POG
alias and the spending amount. I've pulled them together into a single
datasheet that has both COFOG2 and region for each row.

However, about 500 of the rows aren't duplicates, and these are all LA
items - see:

Unmatched rows in Table 9:
https://spreadsheets.google.com/ccc?key=0Atncguwd4yTedHc1Y2xnMEtxOFk3RHQ5S01DS0tmV3c&hl=en&authkey=CPqDpoQC
Unmatched rows in Table 10:
https://spreadsheets.google.com/ccc?key=0Atncguwd4yTedElkUmFfVzUtMi1pZGNhOGV0Q1hOSUE&hl=en&authkey=COK21JgG

(ignore anything that isn't ENG_LA, the others are easily fixed)

Table 9 rows have COFOG2 classifications, as you can see; Table 10 rows don't.

Sometimes you can infer the COFOG2 classifications of the rows in
Table 10, by comparing the two tables (as in the case of 2. Defence),
but most of the time not.

At least, I don't think you can.

>> Actually we should think about how best to add LA spending as unique
>> items... we have the choice of adding these items classified *either* by
>> COFOG 2 or by region, but not both.
>
> Aren't there three dimensions of the data at play
> here?
>
>        entity: local authorities collectively
>        region: one of the NUTS regions
>        function: something from COFOG
>
> Why do we have to lose any of this information?

We don't have to lose COFOG1, but I think we do have to lose the
COFOG2 codes in Table 9, because it's either that or the regional
information.

> If we do have to lose something for some reason,
> I would suggest to leave off the COFOG since that
> can be inferred from the fact that the entity is
> a local authority, whereas it is not possible to
> infer the region from the other information we
> have.

I think that's in agreement with my suggestion above - we should take
the rows from Table 10, so we keep the regional information, and
replace COFOG2 codes with 'local authority'.